David Steensma, MD of the Dana-Faber Cancer Institute, Boston, MA gives an overview of his talk on how to treat patients with difficult forms of myelodysplastic syndromes (MDS) held at the 2016 Annual Meeting of the British Society of Haematology (BSH) and International Society of Hematology (ISH) in Glasgow, Scotland. Dr Steensma discusses how for some of these patients, there is no prescribed algorithm or obvious approach for treatment. This includes lower risk patients for whom erythropoietin and lenalidomide have failed and high-risk patients who have been failed by azacitidine or decitabine, as well as post-transplant relapses and patients whose primary cytopenia is neutropenia or thrombocytopenia rather than anemia. According to Dr Steensma, there is no good data to support decision making for these groups of patients and the focus of the session will be to discuss with other clinicians, what is already know and what the key priority for future studies should be. In terms of what is known, he speaks about the subset of patients who respond to immunosuppresive therapy, those who have a small PNH clone, those that have normal karyotype or tricomy 8 and those patients that are slightly on the younger side of the median age of 70. For example those patients with MDS who are younger than 60, and patients with HLA-DR15 are more likely to respond to immunosuppressive therapy. Patients with excessive blasts and patients with complex karyotypes monosomy 7 are less likely to respond to immunosuppressive therapy. As of now, there is no specific biomarker or signature to indicate the likelihood of response to immunosuppressive therapy for these patients. Similarly for transplants, patients with p53, U2AF1 or NRAS mutations, there is a high risk of relapse. Those with proliferative disease, MDS/MPN overlap or more than 10% blasts, also seem to have a higher risk of relapse. Dr Steensma suggests that we need to find new approaches to treatment of these as well as those patients who seem to be good candidates for transplant as outcomes are still sub optimal.